Climate change and adaptive capacity to extreme events in
the Rhine basin
A.H. te Linde1,2, J.C.J.H. Aerts1 1 Institute for Environmental Studies (IVM - VU), De Boelelaan 1087, 1081 HV Amsterdam; aline.te.linde@ivm.vu.nl 2 Deltares, Rotterdamseweg 185, 2629 HD Delft
Abstract differences in future change of precipitation
Impacts of land use change, flood defense and temperature. Also, the present day measures and climate change on peak variance of those parameters are left discharges are studied at different scales in the unchanged, while changes can be expected Rhine basin. Few studies, though, have due to climate change. To overcome these studied the combined effect of these variables limitations, we applied a method that uses in scenarios and how these scenarios change (bias-corrected) direct output from the regional the return period of flood peaks. In the current climate model (RCM) RACMO (Lenderink, paper we introduce a method to combine 2003), forced by GCM ECHAM 5 as well. simulations of the effect of upstream flooding To estimate extreme discharges at Lobith, and climate change on flood-peak probability. we added an extra shape parameter to the Gumbel distribution, and applied the Weibull Introduction distribution. Recent floods and droughts reinforced cross- boundary cooperation on flood management in the Rhine basin (e.g. IKSR Flood Action Plan, Dutch-German Working Group on Floods (since 1997), EU Floods Directive (2007)). However, these initiatives do not consider the effect of climate change in the Rhine basin, while recent research findings conclude that as a result of climate change peak discharge is likely to move from spring to winter due to early snowmelt in addition to precipitation increase (e.g. Kwadijk and Middelkoop, 1994; Middelkoop et al., 2001; Kleinn et al, 2005). Figure 1. Flow chart displaying all modelling steps Furthermore, historical time series are too short to derive a statistically sound Results extrapolation of return periods of flood peaks with a low probability. In the current paper we Climate change impact introduce a method to simulate low probability Following the climate change scenarios, mean floods, and combine the effect of upstream winter discharge is expected to increase by 11 flooding and different climate change scenarios % according to the Wp scenario and to 14 % on peak discharges. according to the RACMO scenario (Table 1).
Method Table 1: Seasonal change in discharge at Lobith according
All modelling steps are visualized in Fig. 1. We to KNMI’06 scenarios and RACMO output, for 2050 used the semi-distributed conceptual HBV Q (m3/s) DJF MAM JJA SON Year model for rainfall modeling and the Observed 2778 2518 2314 1809 2355 hydrodynamic model SOBEK to recalculate all 1961-1995 yearly maximum discharges. To produce dQ G (%) 6.71 2.03 -2.16 1.47 2.01 dQ Gp (%) 5.58 3.00 -17.40 -17.36 -6.61 discharge series of at least 1000 years we dQ W (%) 13.04 4.47 -3.94 3.02 4.15 applied the stochastic rainfall generator dQ Wp (%) 10.56 6.50 -31.68 -33.25 -11.97 developed for the whole Rhine basin by dQ RACMO 14.20 -0.75 -17.36 -12.63 -4.14 Beersma (2001). (%) We constructed specific climate scenarios for the Rhine basin by applying the delta Figure 2 shows that the spatial variation of the change approach on a historical dataset (1961- projected change differs to a large extent 1995), based on the KNMI’06 scenarios (Te between RACMO and Wp. This is most likely a Linde, 2007). But the delta change does not direct result from the spatial variation in direct take into account possible geographical RCM output, compared to the uniformly spread delta change approach. a)
b)
Figure 2. Mean change in winter discharge for the Wp and
RACMO scenarios (2050)
Interpreting extremes and effect of flooding
According to Figure 3 and Table 2, the flood frequency will increase as a result of climate change. For the situation without flooding, a Figure 3: Extreme value distribution of yearly maximum at 3 Lobith The Weibull distribution fits are shown without peak discharge of 15,000 m /s with a return confidence intervals. Displayed are the reference situation period of 500 years (Gumbel variate is 6.2) in (a) and 2050 ac-cording to the Wp scenario (b) the reference situation will shift to a return period of 100 years (Gumbel variate is 4.6) in the Wp scenario. Also, the 1/1000 year event modelling. This enabled us to simulate low 3 of 15,700 m /s in the reference situation will probability floods, and to combine the analysis 3 shift to 18,200 m /s in 2050. If we do take into on the effect of upstream flooding and different account the simulated effect of flooding, these climate change scenarios on peak discharges. differences in peak discharges and return periods between the reference and the Wp References scenario for 2050 are less dramatic. The Beersma, J.J., Buishand, T.A., Wójcik, R. (2001) Rainfall 1/1000 year event will then shift from 14,000 generator for the Rhine basin; multi-site simulation of 3 3 daily weather variables by nearest-neighbour m /s in the reference situation to 15,000 m /s in resampling. P. Krahe, Herpertz, D. Lelystad, the 2050, which is a reduction of 10 – 15%. Netherlands, International Com-mission for the Hydrology of the Rhine basin CHR. CHR-Report no. I- Table 2 Estimated return periods obtained by ranking the 20: 69-77. peak events at Lobith according to size and linking return Kleinn, J., Frei, C., Gurtz, J., Lüthi, D., Vidale, P., Chär, C. periods to the ranks. A dataset of 1000 years was used (2005) Hydrologic simulations in the Rhine basin driven by a regional climate model. Journal of Return Geophysical Research,110 (D04102, Reference Climate change (Wp) period doi:10.1029/2004JD005143). Without With Without With Kwadijk, J., & Middelkoop, H. (1994) Estimation of impact flooding flooding flooding flooding of climate change on the peak discharge probability of 1000 15,700 14,000 18,200 15,400 the river Rhine. Climatic Change, 27, 199-224. 500 15,000 13,700 17,700 14,800 Lenderink, G., Hurk, B. Van den, Van Meijgaard, E., Van 200 14,300 13,100 16,700 14,500 Ulden, A., & Cuijpers, J. (2003) Simulation of present- 100 12,900 12,600 15,200 13,600 day climate in RACMO2: first results and model developments. (Tech. Rep. No. 252). KNMI. Te Linde, A.H. (2007) Effect of climate change on the Conclusion discharge of the rivers Rhine and Meuse. Applying the Results show that the flood-peak probabilities KNMI 2006 scenarios using the HBV model. Delft, The are likely to increase as a result of climate Netherlands, WL | Delft Hydraulics: 1-30. change in the Rhine basin, and that there are Middelkoop, H., Daamen, K., Gellens, D., Grabs, W., Kwadijk, J.C.J., Lang, H., Parmet, B.W.A.H., Schädler, large (spatial) differences between climate B., Schulla, J., Wilke, K. (2001) Impact of climate change scenarios. The simulated effect of change on hydrologi-cal regimes and water resources flooding decreases peak discharge at Lobith by management in the Rhine basin. Climatic Change 49: 105-128. 10 – 15%. We used a modelling method that combines a weather generator and different climate change scenarios with hydrological